Interesting point about energy cost. If GPT inference keeps scaling, latency + watt efficiency might become central to AGI deployment.
Interesting point about energy cost. If GPT inference keeps scaling, latency + watt efficiency might become central to AGI deployment.